ECS K7S6A with SiS745 in the test: The successor to the SiS735 is here

ECS K7S6A with SiS745 in the test: The successor to the SiS735 is here


Hardly any motherboard has attracted such crowds in recent months Aryanized, like the ECS K7S5A. The SiS735 used at the time offered very good performance at an almost unrivaled price. But the board did extraordinarily bravely in our review, at the latest a few weeks after the market launch the forums with threads on this board were just overflowing. Incompatibilities with exotic hardware, no options for overclocking and a whole series of inexplicable quirks made it difficult for the buyer. In this article we want to take a closer look at the quasi-successor K7S6A. By using the SiS745, the board promises support for PC2700 (DDR333) RAM at another low price. According to the motto 'two birds with one stone' we will take a look at a sample example of the currently available DDR333 memory. Because the modules manufactured without any real JEDEC specification have anything but a good reputation.

The chipset: SiS745

The evolutionary step that SiS has taken between SiS735 and 745 is not great. In addition to the support of DDR333 Ram as well as the omission of SD-RAM support and the integration of a FireWire controller (which is not used by the K7S6A), no new features have migrated into the chip.

Another prominent feature of the Chipset is likely to be the 'one-chip design'. Since SiS was able to combine all functions in one module, the otherwise usual division into north and south bridges is no longer necessary. Since the chip has the same properties as its predecessor, we refer to our SiS735 Review .

The SiS745

Sense or nonsense: DDR333?

DDR266 or DDR333? 2.1 or 2.7 GByte data transfer rate per second? Where on paper a theoretical increase in performance of 29% seems to indicate considerable speed advantages, in practice two main problems of the ideal world stand in the way. Damper number one is due to the architecture of current CPUs and mainboards, which will still rely on an FSB of 133MHz in the foreseeable future and thus can still only record 2.1GByte per second. The theoretical 2.7GB of memory can therefore only be used indirectly if several components really do argue about the bandwidth. In this case the processor should have more reserves than with DDR266. The synchronous operation with FSB166 should bring a real advantage.

The second problem is that currently none Uniform JEDEC specification for DDR333 RAM and the required module boards exist. While the memory chips are already doing their job to a certain extent with a clock rate of 166MHz, the 'old' ones take care of themManufacturers drilled out circuit boards on their own for a lot of trouble. The signal noise is often fatal even with the slightest tightening of the timing, the computer no longer boots or crashes suddenly. It should quickly become clear that the performance of these currently available CL2.5 modules cannot exactly reach new heights. Just like the fact that Infineon can and will only deliver its first DDR333 modules in a size of 128MB. No seal of approval for the 256MB and 512MB bars currently floating around on the market. In our review we will try to clarify whether the PC2700 can at all set itself apart from its 'predecessor' under the current circumstances.

(update 28.03.2002) : The JEDEC now has finally the pending specifications for DDR333 passed . Thus, in the coming weeks, you can finally expect a high-quality PC2700.

On the next page: Scope of delivery and documentation